Calculating models fairness.

I am going to calculate models fairness. I use adult income dataset. Models that I use are Random Forest Regressor and Logistic Regression.

First model that I use is RandomForestClassifier. I have already used RandomForestRegressor before. Difference is that Classifier outputs probability distribution over classes, not a single number. Later I apply bias mitigation methode 'roc-pivot' to the explainer.

Logistic regression aims to solve classification problems. It does this by predicting categorical outcomes, unlike linear regression that predicts a continuous outcome.

  1. These coefficients differ between models, we can see many too low values for first one while same coefficients are too high or match norms in a second model.
  2. After applying roc-pivot, there is no improvement, I have tried also with reweight, but it did not work either.
  3. Based on these models I do not see correletion, all models are highly unfair.